ASTRONOMY
AND
ASTROPHYSICS
Detection of HD in the Orion molecular outflow
?
Frank Bertoldi1,2, Ralf Timmermann2, Dirk Rosenthal2, Siegfried Drapatz2, and Christopher M. Wright3,4 1 Max-Planck-Institut f¨ur Radioastronomie, Auf dem H¨ugel 69, D-53121 Bonn, Germany (bertoldi@mpifr-bonn.mpg.de) 2 Max-Planck-Institut f¨ur Extraterrestrische Physik, Giessenbachstrasse, D-85740 Garching, Germany
3 Leiden Observatory, P.O. Box 9513, 2300 RA Leiden, The Netherlands
4 School of Physics, University College, UNSW, ADFA, Canberra ACT 2600, Australia (wright@ph.adfa.edu.au)
Received 5 November 1998 / Accepted 18 February 1999
Abstract. We report a detection in the interstellar medium
of an infrared transition within the electronic ground state of the deuterated hydrogen molecule, HD. Through a deep in-tegration with the Short-Wavelength-Spectrometer (SWS) on board the Infrared Space Observatory (ISO), the pure rotational v = 0 − 0 R(5) line at 19.43 µm was detected toward the Orion (OMC-1) outflow at its brightest H2 emission region, Peak 1. The∼ 2000beam-averaged observed flux of the line is
(1.84 ± 0.4) × 10−5erg cm−2s−1sr−1. Upper flux limits were
derived for sixteen other rotational and ro-vibrational HD lines in the wavelength range 2.5 to 38µm.
We utilize the rich spectrum of H2 lines observed at the same position to correct for extinction, and to derive a total warm HD column density under the assumption that similar excitation conditions apply to H2and HD. Because the observed HD level population is not thermalized at the densities prevailing in the emitting region, the total HD column density is sensitive to the assumed gas density, temperature, and dissociation fraction. Ac-counting for non-LTE HD level populations in a partially disso-ciated gas, our best estimate for the total warm HD column den-sity isN(HD) = (2.0 ± 0.75) × 1016cm−2. The warm molec-ular hydrogen column density is(2.21±0.24)×1021cm−2, so that the relative abundance is[HD]/[H2] = (9.0 ± 3.5) × 10−6. The observed emission presumably arises in the warm lay-ers of partially dissociative magnetic shocks, where HD can be depleted relative to H2due to an asymmetry in the deuterium-hydrogen exchange reaction. This leads to an average HD de-pletion relative to H2of about 40%. Correcting for this chem-ical depletion, we derive a deuterium abundance in the warm shocked gas, [D]/[H]=(7.6 ± 2.9) ×10−6.
The derived deuterium abundance is not very sensitive to the dissociation fraction in the emitting region, since both the non-LTE and the chemical depletion corrections act in oppositite direction. Our implied deuterium abundance is low compared to previous determinations in the local interstellar medium, but
Send offprint requests to: F. Bertoldi
? Based on observations with ISO, an ESA project with instruments funded by ESA Member States (especially the PI countries: France, Germany, The Netherlands and the United Kingdom) and with the participation of ISAS and NASA.
it is consistent with two other recent observations toward Orion, suggesting that deuterium may be significantly depleted there.
Key words: shock waves – ISM: abundances – ISM: individual
objects: Orion Peak 1 – ISM: molecules – cosmology: observa-tions – infrared: ISM: lines and bands
1. Introduction
Deuterium is an important clue to the physics of the Big Bang. Its creation rate during primordial nucleosynthesis depended strongly on the number ratio of photons to baryons, a quan-tity not at all well-known but crucial for a correct description of the earliest events (e.g., Wilson & Rood 1994; Smith et al. 1993). Ever since, conditions have not been right to add further deuterium to the primordial production; neither nuclear fusion processes nor the spallation of heavier nuclei by energetic cos-mic rays can augment the original abundance (although stellar flares were suggested by Mullan & Linsky [1999] to produce deuterium). The deuterium abundance in fact decreases contin-uously as deuterium is burned up in stars. The present day deu-terium abundance in the interstellar medium provides a lower limit to its primordial value, and it reflects the history of stellar reprocessing of the gas; measurements of its spatial variation may shed light on the star formation history in a given region (e.g., Tosi 1998, Tosi et al. 1998).
performed by Jenkins et al. (1999) and Sonneborn et al. (in prep.) along three lines of sight, which yield significant [D]/[H] variations between7.4 × 10−6 and2.1 × 10−5. Observations of the hyperfine transitions of Di and H i at radio frequencies (Chengalur et al. 1997), or of the H2and HD molecules, pro-vide further means to sample the deuterium abundance. The [HD]/[H2] ratio is subject to variations due to small differences in chemical reaction rates in high temperature molecular gas as in shocks (see Sect. 3.5) and photodissociation regions. To date, rotationally excited HD has been detected with ISO from giant planets (Feuchtgruber et al. 1997), and in the interstellar medium through UV-observations by Wright & Morton (1979). Accounting for the three lowest rotational levels of HD Wright & Morton found [HD]/[H2]< 6 × 10−7in the cold molecular gas towardζ Oph. D-bearing species such as HDO, CH2DOH, DCO+ or DCN (Jacq et al. in prep.), are observed at radio-frequencies in the ISM, but due to fractionation the deuterium abundance cannot be deduced accurately from these observa-tions.
The infrared emissivity of HD has been modeled and pre-dicted for photodissociation regions (Sternberg 1990 - not in-cluding the reactionH+HD ←→ H2+D) and magnetic (C-type) shocks (Timmermann 1996), where latter calculations include deuterated species in the chemical reaction network. But even state-of-the-art IR-spectrometers had been unsuccessful in de-tecting HD emission in the ISM, because of the low HD abun-dance, and because the pure rotational and rotation-vibrational lines appear at wavelengths that are affected by strong telluric interference.
ISO now for the first time opened the skies to a successful search for HD emission. Wright et al. (1999) detected the emis-sion of HD 0-0 R(0) toward the Orion Bar photodissociation region, and derive [HD]/[H2]= (2.0 ± 0.6) × 10−5. A measure of the deuterium abundance in PDRs is complicated by the fact that the HD dissociation front is located deeper into the molec-ular cloud than that of H2. The average excitation of HD may therefore be lower than that of H2, making it difficult to derive column densities referring to the same regions. In regions of pure collisional excitation such as shocks, this problem should not occur.
We here report HD observations toward the brightest H2 emission region in the sky, Peak 1 in the Orion molecular outflow that surrounds several deeply embedded far-infrared sources (Genzel & Stutzki 1989; Menten & Reid 1995; Blake 1997; van Dishoeck et al. 1998; Stolovy et al. 1998; Schultz et al. 1998). Our observations, which we discuss in Sect. 2, were part of a line survey of shocked molecular gas, and of a program to in-vestigate the oxygen-chemistry in the warm molecular interstel-lar medium with the Short-Wavelength-Spectrometer (SWS). In Sect. 3 we derive an HD column density from the flux of one detected HD line, making use also of extinction and excitation measurements from a large number of H2lines which we de-tected in related observations. We then estimate the deuterium abundance in the shocked, warm gas of the Orion outflow. In Sect. 4 we summarize our results.
2. Observations
Orion Peak 1 was observed in the SWS 01 (full grating scan) and SWS 07 (Fabry-P´erot) modes of the short wavelength spec-trometer (de Graauw et al. 1996) on board ISO (Kessler et al. 1996) on October 3, 1997, and in the SWS 02 (≈ 0.01 λ range grating scan) mode on September 20, 1997 and February 15, 1998. Fig. 1 illustrates the various aperture orientations with re-spect to the H21–0 S(1) emission observed with NICMOS on the HST (Schultz et al. 1998). The full 2.3 to 45µm SWS 01 spectrum was recorded in its slowest mode with the highest pos-sible resolution. A preliminary reduction of this spectrum was presented by Bertoldi (1997). Table 1 summarizes the HD line observations. The H2lines will be discussed in more detail in a forthcoming article (Rosenthal et al. in prep.).
The data reduction was carried out using standard Off Line Processing (OLP) routines up to the Standard Processed Data (SPD) stage within the SWS Interactive Analysis (IA) system. Between the SPD and Auto Analysis Result (AAR) stages, a combination of standard OLP and in-house routines were used to extract the individual spectra. The in-house routines included an interactive dark-current subtraction for individual scans and detectors as well as for the removal of fringes. The flux cal-ibration errors range from 5% at 2.4 µm to 30% at 45 µm. (SWS-Instrument Data User Manual, version 3.1). The statisti-cal uncertainties derived from the line signal to noise ratio are for most detected lines smaller than the systematic errors due to flux calibration uncertainties.
2.1. Line fluxes
The spectra at the expected wavelengths of seventeen HD transi-tion lines are shown in Figs. 2 and 3. The wavelengths (Table 1) of thev = 0 − 0 pure rotational HD transitions were adopted from Ulivi et al. (1991), while those for the rotation-vibrational lines were computed from the rotational (Essenwanger & Gush 1984) and vibrational constants (Herzberg 1950). Only one of the seventeen lines sought was detected, the 0–0 R(5) (i.e. J = 6 → 5) dipole transition at an expected wavelength of 19.4305µm1 (observed at 19.4290 µm, see Fig. 3). Although
the detection may seem marginal, the line does appear in two independent observations. It is apparent in the fringed data, and after defringing, it stands out well above the noise in the cen-tral wavelength range of the coadded SWS 02 scans. A beam-averaged flux of(1.84±0.4)×10−5erg s−1cm−2sr−1was de-rived from a continuum-subtracted integration over the feature; the error derives from the line’s S/N= 4.5 (the RMS noise be-ing evaluated within±500 km s−1) plus an estimated 11% flux calibration uncertainty. The line width (FWHM) is 134 km s−1, which agrees with the instrumentally expected width for ex-tended objects, 130 km s−1, and is not sufficient to resolve the emission, which should have a velocity dispersion similar to that
1
The experimental value for the R(5) transition wavelength is
19.4305 ± 0.0001 µm, whilst the calculated value is 19.431002 ± 0.000008 µm (Ulivi et al. 1991 and references therein). The
32
*
*
*
Peak1
IRc2
Peak2
50
49
48
47
46
45
44
05
-5 25 00
45
30
15
24 00
45
30
Right Ascension (1950)
Declination (1950)
h m s s s s s s s ˚ ' " " " " " " " ' SWS02 (September 20, 1997) SWS01 (October 3,1997 ) SWS02 (February 15, 1998)BN
SWS07 (October 3,1997 )*
Fig. 1. The various SWS apertures of our
ISO observations of the Orion outflow, over-laid on a NICMOS H21–0 S(1) map kindly provided by E. Erickson and A. Schultz (Schultz et al. 1998). White dots and the central white patch are continuum ghost-images of stars and of Orion-BN, respec-tively. The ISO aperture was centered on
05h32m46s.27, −05◦ 240 0200(1950). Its size corresponds to1400× 2000 for wave-lengths smaller 12µm, 1400× 2700at 12 to 27.5µm, 2000× 2700at 27.5 to 29 µm, and
2000× 3300at 29 to 45.2 µm.
Table 1. Summary of ISO-SWS observations of HD lines at Orion Peak 1.
transition λa integrationb peak ∆vc A EvJ Iobs Aλ NvJd
µm sec Jy km s−1 s−1 K erg s−1cm−2sr−1 mag cm−2
0–0 R(2) 37.7015 < 192 300 1.72(–6)e 765.9 <9.9(–4) 0.09 < 1.5(17) 0–0 R(3) 28.5020 < 100 350 4.11(–6) 1270.7 < 7.9(–4) 0.30 < 4.6(16) 0–0 R(4) 23.0338 < 35 350 7.91(–6) 1895.3 < 6.0(–4) 0.51 < 1.8(16) 0–0 R(5) 19.4305 3010 2.37 134 1.33(–5) 2635.8 (1.84 ± 0.4)(–5) 0.60 (3.0 ± 1.1)(14) 0–0 R(6) 16.8940 < 8.5 230 2.03(–5) 3487.5 < 1.3(–4) 0.53 < 1.1(15) 0–0 R(7) 15.2510 1110 < 5.0 185 2.88(–5) 4445.3 < 6.8(–5) 0.41 < 3.3(14) 0–0 R(8) 13.5927 < 10 215 3.87(–5) 5503.8 < 1.8(–4) 0.37 < 5.6(14) 0–0 R(9) 12.4718 < 8 230 4.97(–5) 6657.5 < 1.7(–4) 0.54 < 4.3(14) 1–0 P(4) 3.0690 < 0.2 280 7.37(–6) 5958.5 < 2.8(–5) 1.11 < 2.0(14) 1–0 P(3) 2.9800 < 0.1 270 1.09(–5) 5593.5 < 1.4(–5) 1.08 < 6.4(13) 1–0 P(2) 2.8982 < 0.25 300 1.64(–5) 5348.7 < 3.9(–5) 0.94 < 1.0(14) 1–0 P(1) 2.8225 < 0.4 305 3.23(–5) 5225.7 < 6.6(–5) 0.80 < 7.6(13) 1–0 R(0) 2.6900 < 0.2 330 1.72(–5) 5348.7 < 3.7(–5) 0.70 < 7.0(13) 1–0 R(1) 2.6326 < 0.3 330 2.51(–5) 5593.5 < 5.7(–5) 0.70 < 7.3(13) 1–0 R(2) 2.5811 < 0.4 245 3.22(–5) 5958.5 < 5.8(–5) 0.72 < 5.7(13) 1–0 R(3) 2.5350 1200 < 0.2 215 3.91(–5) 6441.1 < 2.6(–5) 0.74 < 2.1(13) 1–0 R(4) 2.4943 < 0.35 235 4.60(–5) 7037.7 < 5.0(–5) 0.76 < 3.5(13) aexpected wavelengths. 0-0 R(5) is observed at19.4290 µm.
bon target integration time for SWS 02 observations; the SWS 01 2.4–40 µm scan took 6538 sec. cFWHM of observed HD line, or of neighboring H
2lines. dupper level column, corrected for extinction.
Fig. 2. Spectra of HD non-detections. Three of the spectra were obtained in the SWS 02 mode with ISO, all others were taken in the SWS 01
mode. Neighboring H2and HI lines are identified at their expected positions.
observed in H21–0 S(1), which has a FWHM of≈ 50 km s−1, with emission from -100 to +100 km s−1(Chrysostomou et al. 1997; Stolovy et al. 1998). The line center is positioned at vhelio ' −23 km s−1, which is within the range of line
cen-ter velocities,vhelio' −38 to +41 km s−1, observed in the H2 1–0 S(1) emission (Chrysostomou et al. 1997). The high spec-tral resolution and sensitivity of the ISO SWS around 19µm made0 − 0 R(5) the most promising line for a detection of HD in Peak 1. The stronger lines from lower rotational states suffer from the rapidly rising continuum level at longer wavelengths and the resulting strong fringing.
The available integration time was insufficient to detect two other HD lines which we tried to observe in the SWS 02 mode. We derived upper flux limits for a total of sixteen HD lines from the noise level near the expected line wavelengths (Table 1).
The peak-to-peak noise envelope approximately corresponds to a 3σ flux density dispersion, and we assumed that a line with a peak flux density of 3σ would have stood out clearly enough to be detected. As upper flux limits we therefore adopted the 3σ flux density noise level times the FWHM of neighboring H2 lines, except for 0–0 R(7), R(8), R(9), for which we adopted the linewidths expected from instrumental resolution.
3. Extinction correction and H2excitation
The column density of molecules in a particular rotation-vibrational level(v, J) is computed from the observed line flux, Iobs(v, J → v0, J0), of a transition to a lower state (v0, J0), times
(Ein-Fig. 3. Defringed AOT 02 spectra of two 0–0 R(5) observations
per-formed at the labeled dates, and the merged spectrum. The line is visible even in the original fringed data of both observations. Continuum levels differed in both datasets (due to differing orientations of the aperture with respect to Orion-BN), so that a constant was added to one before averaging them, and in order to display them together. The rising noise near the scan edges is expected and due to the lower detector coverage. In the merged spectrum, the line stands out with a peak flux density to RMS noise ratio of 4.5.vheliorefers to the expected line wavelength of 19.4305µm.
stein) coefficient2for this transition,A(v, J → v0, J0), times a correction for extinction along the line of sight:
NvJ = 4πhc λ Iobs(v, J → v 0, J0)
A(v, J → v0, J0) 100.4Aλ, (1)
whereAλis the effective extinction at wavelengthλ. All H2and HD transitions are optically thin. Previous measurements of the Peak 1 molecular emission estimated K band (2.12µm) extinc-tions between 0.5 and 1 magnitude (Everette et al. 1995). To apply an extinction correction to the observed H2and HD lines requires knowledge of the “extinction law”Aλ/AKbetween 2.4 and 40µm, which especially above ∼ 5 µm is observationally not well constrained (Draine 1989). The near-IR extinction is usually approximated to follow a power law (see Fig. 4), Aλ= AK (λ / 2.12 µm )− for λ < 6 µm, (2)
2 The Einstein coefficients for H
2 and HD are taken from Turner et al. (1977), confirmed by Wolniewicz et al. (1998), and Abgrall et al. (1982), respectively. The transition energies for H2were computed from the level energies that were kindly provided by Roueff (1992, private communication).
with an increase in opacity beyond 6µm due to stretch and bend mode resonances in silicate grains (Draine & Lee 1984), the shape and depth of which are yet poorly understood. A recent ISO study of the extinction toward the W51 Hii region, e.g., findsA19/A10' 0.52±0.1 and A19/AK' 0.57±0.1 (Bertoldi et al. in prep.). Draine & Lee (1984) suggested A18/A9.7 '
0.40, and observations of circumstellar dust emissivities suggest
values for this ratio between 0.35 (Pegourie & Papoular 1985) and 0.5 (Volk & Kwok 1988).
The effective extinction toward the emitting gas in Peak 1 may not necessarily follow an average interstellar extinction law, since the emitting and absorbing gas may be mixed. We therefore tried to estimate the effective extinction as a function of wavelength toward Peak 1 from the observed emission of Peak 1 itself. The differential extinction between two wavelengths can be derived from a comparison of H2 line fluxes of transitions arising either from the same upper level, or from neighboring thermalized levels with a well determined relative excitation. Since the warm HD and H2are likely to be well mixed, we will use the extinction toward the H2to deredden also the HD line intensities.
The extinction-corrected H2 column density distribution serves as a “thermometer” that probes the excitation conditions in the emitting gas. Since we detected only one HD line, we can-not determine the HD excitation directly –this would require at least two lines of transitions arising from different upper levels. Instead, we rely on the reasonable assumption that HD is subject to the same excitation conditions as H2. We can thereby derive both molecules’ total column densities in the warm, emitting gas, and from that, their abundance ratio.
3.1. 2.4–6µm extinction
It is useful to plot the observed H2 or HD column densities, NvJ, divided by the state’s statistical weight,3gJ, against the state’s energy,EvJ. Examining this “excitation diagram,” which we first plotted from the not yet extinction-corrected H2 line intensities, we found no indication up toEvJ/k ≈ 40, 000 K of fluorescent excitation, or for deviations from the statistical ortho- to para-H2abundance ratio of 3 (details will be given by Rosenthal et al. in prep.). Instead, the populations appear both in thermodynamic and statistical equilibrium, which means, they depend only on the level energy and degeneracy, not on the vibrational quantum number. As a consequence, NvJ/gJ is a smooth function of the level energy.
In contrast, H2 that is fluorescently excited generally dis-plays level populations that are not in vibrational and rotational LTE, and also show ortho-to-para column density ratios between neighboring states that are smaller than 3 (see e.g., Draine & Bertoldi 1996, Bertoldi 1997). Timmermann (1998) predicted that in low-velocity shocks the ortho-to-para ratio can also be
3 The statistical weights areg
J = 2J + 1 for para-H2(evenJ) and
Fig. 4. Adopted extinction curve (solid line), including water ice and
silicate features. Dashed lines show curves with different relative sil-icate features strengths,A19/A10 = 0.35 and 0.52. The dash-dotted curve assumes narrower 9.7µm feature. Wavelenghts of those H2lines used to constrain the extinction curve are marked – but these are not datapoints!
significantly lower, which was confirmed in recent ISO-SWS observations by Neufeld et al. (1998), who found a value of ' 1.2 in the outflow HH54. Peak 1 however shows no devia-tions from statistical equilibrium.
We were able to obtain the most reliable line fluxes for the H2v = 0−0 and v = 1−0 lines between 2.4 and 6 µm that have upper level energiesEvJ/k = 5000–16, 000 K. Excluding lines in the water ice absorption feature between 2.8 and 3.3µm, we selected thirty-two high S/N lines in this wavelength and energy range to constrain the power law part of the extinction curve. Assuming that the H2 columns NvJ/gJ vary smoothly with EvJ, we fit a second order polynomial to the observed column densities, corrected for extinction following Eq. (1) with AK and as free parameters. We searched for values of AK and which minimize the dispersion of NvJ/gJ. Within the range = 1−2, AKis well constrained to1.0±0.1 magnitudes. It was
not possible to narrow down the value of further, so we adopted = 1.7, which was suggested by previous observational studies (see Draine 1989; Brand et al. 1988 usedAK= 0.8 and = 1.5). The second order fit toNvJ/gJversusEvJ/k ≡ 1000 T3K for AK= 1.0 and = 1.7 is (see Fig. 5)
log(NvJ/gJ) = 18.88–0.402 T3+ 0.0092 T32, (3)
for5800 K < EvJ/k < 17, 000 K, and has a fit quality χ2 =
0.13; the uncorrected (AK= 0) dispersion is χ2= 0.39.
We should note that our extinction curve between 4 and 7 µm is not constrained well enough to address the claim by Lutz et al. (1997) that the extinction curve flattens and lacks the presumed minimum at 7 µm.
H
2in ORION Peak1
0 5000 10000 15000 EvJ/k [K] 1014 1015 1016 1017 1018 1019 1020 NvJ /gvJ [cm -2 ] v=0, JU= 2 3 4 5 6 7 8 9 10 11 12 13 14 15 v=1 JU=012 3 4 5 6 7 8 9 10 11 v=2 JU=012 3 4 5 6 7 8Fig. 5. H2excitation diagram: the dots, diamonds, and squares denote dereddened and beam-averaged column densities of transitions in the
v = 0–0, 1–0, and 2–1 bands, respectively. The solid line represent the
least squares fit, Eqs. (3) and (4), which apply forEvJ/k larger and smaller than 5800 K, respectively.
3.2. Mid-IR extinction
be-tween two measurements of the 0–0 S(3) line in different AOTs. When we narrow the 10µm feature below reasonable width es-timates, the extinction correction for the J = 4 and J = 6 levels decreases significantly, thereby lowering the dereddened flux in the 0–0 S(3) line and decreasing the required extinction toA9.7≈ (1.1–1.3) mag.
Considering all uncertainties we estimateA9.7 ≈ (1.35 ±
0.15) mag, and with A19/A10 ≈ 0.35–0.52, we find A19 ≈ (0.61 ± 0.15) mag. Our value for A19/AK= (0.61 ± 0.15) is
in good agreement with the0.57 ± 0.1 found toward W51. The extinction correction for the detected HD line we then estimate as100.4A19 ≈ 1.75 ± 0.25. The dereddened column
densities of the H2v = 0, J ≤ 8 levels are fit by
log (NvJ/gJ) = 20.17 − 0.765 T3+ 0.0344 T32. (4)
At upper level energiesEJv/k = 0, 2636 K (HD J = 6) and 5800 K this corresponds to excitation temperaturesTex' 570, 740 and 1190 K, respectively.
3.3. Water ice feature
The rotation-vibrational HD transitions 1–0 P(2), P(3), and P(4) fall into the water ice absorption feature at3.05 ± 0.25 µm. To correct these line flux limits for extinction we estimated the depth and width of the feature from the apparently enhanced (over the power law, Eq. [3]) extinction of five H2 lines with
6000 K < Eu/k < 16, 000 K and 2.8 µm < λ < 3.3 µm. The Gaussian ∆Aice(λ) ≈ 0.58 e−[(λ−3.05µm)/( √ 2·0.15µm)]2 mag (5)
approximately fits the additional extinction noticeable for the five H2lines, and was used to correct the rovibrational HD line fluxes.
3.4. Total column ofH2and HD
How can we derive the total HD column density with only one measured high-excitation level? We can make use of the ob-served H2excitation, and estimate that the excitation conditions – i.e., the fractions of the total gas column density that are at a particular temperature and density – are the same for HD and for H2. We assume that at least up to the excitation energy of HD J = 6 (E06/k = 2636 K), the H2level populations are ther-malized, and thereby reflect the kinetic temperature distribution of the gas. The assumption of thermalized level populations is supported by the lack of deviations from vibrational degeneracy or non-statistical ortho-to-para ratios, and by detailed non-LTE calculations of the H2level populations, which show that de-viations from LTE are small atJ ≤ 8 (E/k ≤ 5800 K) for densities higher than105cm−3and temperatures above 600 K (Draine & Bertoldi, unpublished).
Because they are permitted dipole transitions, the radiative decay rates of HD are much higher than those of H2 levels at comparable energy. This results in “critical” densities (i.e. den-sities above which a level is thermalized for a given temperature) which are higher for HD than for H2. If the gas density was high
HD in ORION Peak1
0 2000 4000 6000 8000 EvJ/k [K] 1011 1012 1013 1014 1015 1016 1017 NvJ /gvJ [cm -2 ] v=0 JU= 3 4 5 6 7 8 9 10 v=1 JU=01 2 3 4 5Fig. 6. HD excitation diagram. Pure rotational transitions are denoted
by dots, andv = 1–0 transitions by diamonds. The line represents the fit Eq. (6). The error of the 0–0 R(5) line is computed from spectral noise (22%) and uncertainties in the flux calibration (11%) and the extinction at 19.4µm ('14%).
enough to thermalize the HD levels, then HD would show the same level excitation as H2: normalizing the H2populations (4) with the measured HDJ = 6 level, we would then expect the HD populations to follow
log(NvJ/gJ) = 15.14 − 0.765 T3+ 0.0344 T32 (6)
forT < 6000 K. The measured upper limits to sixteen HD level column densities are consistent with this distribution (Fig. 6).
However, deviations from LTE are important at the expected density of105–106cm−3and temperatures of600–1000 K in the shocked gas of the OMC-1 outflow. LetnvJ/nLTE,vJbe the actual non-LTE population in level(v, J), divided by its LTE value. The total warm HD column density is given by the sum over all level column densities
N(HD) = nLTE,06n 06 X vJ NvJ gJ (2J + 1)nnvJ LTE,vJ, (7)
Fig. 7. Deviations from LTE HD level populations plotted against
the rotation quantum numberJ for different densities and kinetic gas temperatures.n0JandnLTE,0Jrefer to densities in an individualv = 0 state under non-LTE and LTE conditions.
To compute the deviations from LTE of the HD level population we solved the equations of statistical equilibrium (see Timmermann 1996), including radiative decay and exci-tation/deexcitation through collisions with He, H2, and H. The abundance of H varies throughout the shock, and we adopted [H]/[H2]= 1, which is typical for the dissociation fraction in the hot layers of a partially dissociative shock. In Fig. 7 we compare the resulting level population to that in LTE for gas volume den-sities of3 × 105and3 × 106cm−3, and three different kinetic temperatures that should span the range typical of the observed, shocked gas.
HD collisional (de)excitation rate coefficients with H2were computed properly only for pure rotational transitions up to J = 4 and temperatures up to 600 K and 300 K, respectively (Sch¨afer 1990). The rate coefficients for higher level transi-tions and temperatures were extrapolated from the lower ones. The H–HD and H2–HD collision rate coefficients were recently computed by Roueff & Flower (1999) and Roueff & Zeippen (1999), respectively.
We find that for H2densities below106cm−3the HD level populations deviate significantly from LTE. At a density of3 ×
105 cm−3 and a temperature of 600 K, the population of the
J = 6 level is about a factor two below its LTE value.
To assess the non-LTE effects on the derived total HD col-umn density, we evaluatedN(HD) from Eq. (7) and plot this in Fig. 8 as a function of the gas density for three different
temper-Fig. 8. Total HD column density computed from non-LTE level
popu-lations as a function of H2density for three different gas temperatures. The H2density in the preshock gas has been estimated at(1–3.5) ×
105cm−3, and the shocked gas temperatures at≈ 600–1000 K, so we estimateN(HD) ≈ (2.0 ± 0.75) × 1016cm−2.
atures. Only at H2densities above106cm−3do the populations assume LTE, and the warm HD column is
N(HD)LTE = (1.36 ± 0.38) × 1016cm−2. (8)
The error derives from the uncertainty of theJ = 6 column measured by the 0–0 R(5) line, and is a combination of spectral noise (22%), flux calibration uncertainty (11%), and the uncer-tainty of the extinction correction (14%). At lower densities, the total column density much depends on the gas temperature, and since H is the strongest collision partner of HD, on the dissociation fraction.
The emission of the OMC-1 outflow had been modeled previously with C-shocks that propagate at velocities of order
35–40 km s−1 into gas with densities ofn(H
2) = (1–3.5) × 105cm−3(Draine & Roberge 1982; Chernoff et al. 1982;
Kauf-man & Neufeld 1996). The gas in such shocks reaches temper-atures of order 1000 K, and remains at approximately constant density through the region where most of the H2emission oc-curs. The lowest H2 levels we observed show excitation tem-peratures of 600–700 K, which probably reflects the kinetic temperature of much of the warm, emitting gas. Taking these temperatures and densities as estimates for the prevailing excita-tion condiexcita-tions, we estimate that the total observed HD column density must be in the range
N(HD) = (2.0 ± 0.75) × 1016cm−2. (9)
u
s= 35 kms
-1, n
H= 5*10
5cm
-3 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Distance [1015 cm] 100 1000 Temperature [K] Tneutral 0.0 0.5 1.0 1.5 2.0 2.5 3.0 102 103 104 105 106 107 Density [cm -3 ] H2 H HD / 3*10-5Fig. 9. Example of the density and temperature structure in a planar
magnetic C-type shock, which was modeled with the code of Timmer-mann (1996, 1998). The pre-shock densitiesn(H2) = 2.5×105cm−3 and n(HD) = 7.5 cm−3, the shock speed us = 35 km s−1, and
B0 = 700 µG. The depletion of HD relative to H2 reaches a maxi-mum value of 2.4 behind the hottest layer of the shock.
We now compare the HD column to that of warm H2. Sum-ming over the H2level populations given by the least squares fits, Eqs. (3) and (4), we find
N(H2) = (2.21 ± 0.24) × 1021cm−2, (10) where the error reflects a maximum flux calibration uncertainty of 11% in the 7 to 19.5µm range. By summing from J = 0, we extrapolated the observed H2level populations,J ≥ 3, to the unobserved levelsJ = 0–2. We thereby account only for the
warm H2, not for the total H2column along the line of sight, which includes over1022 cm−2 of cold gas in the molecular cloud that embeds the outflow. In this cold gas, which we are not concerned with, most H2is in its ground statesJ = 0 and J = 1, and does not affect our analysis of the warm outflow gas seen in the higher levels.
Dividing the HD and H2column densities, we derive a first estimate of the abundance ratio
[HD]/[H2] = (9.0 ± 3.5) × 10−6. (11) 3.5. Chemical depletion of HD in shocks
The OMC-1 outflow emission arises from warm molecular gas that is shock-heated by high-velocity ejecta which originates
from one of the deeply embedded protostars in the vicinity of Orion IRc2. The emission is a mixture from fast, dissociative J-type shocks, in which the molecular emission comes from where molecules reform, and from slower (< 50 km s−1), partially-or non-dissociative C-shocks, in which the molecules radiate in a magnetic precursor where the temperature rises to its peak value (Fig. 9). In J-shocks that propagate into a medium with density of order105–106cm−3, the fraction of the bulk motion energy radiated away in H2 lines is very small, of order 0.1% (Neufeld & Dalgarno 1989). In C-type shocks, however, about half the kinetic energy is converted to H2emission. For J-type shocks to dominate the H2and HD emission, about 1000 times more energy would have to be dissipated in fast J-shocks than in the slower C-shocks. For an even distribution of magnetic field strengths and shock velocities up to∼ 100 km s−1, C-shocks therefore vastly dominate the molecular emission. The outflow emission may therefore be well modeled as arising from such magnetic, partially-dissociative shocks.
Most of the deuterium is locked in HD in the dense pre-shock molecular gas. In a high-velocity C-pre-shock, both HD and H2are either dissociated through collisions with ions that stream through the neutral gas at a speed of order the shock velocity, or through collisions with warm H2and H at temperatures in excess of 2500 K. In a partly dissociative C-shock, HD is depleted more than H2 because atomic hydrogen can efficiently destroy HD through the reaction
D + H2←→ HD + H + ∆H0, (12)
where∆H0/k = 418 K is the enthalpy difference. Fig. 9 illus-trates the density profiles for H2and HD across such a C-shock. In the warmest layers, HD is depleted by up to a factor 2.4 relative to H2. In the cooling region, the atomic deuterium re-acts stronger again with H2to form HD, and the equilibrium of Eq. (12) is shifted back towards HD. Averaged over the region where the temperature exceeds 400 K, from where most of the observable emission arises, HD is depleted relative to H2by a factor 1.67 in the particular case we display in Fig. 9.
In C-shocks propagating at velocities below 25 km s−1, H2is not dissociated significantly, and because of the low abundance of atomic hydrogen, the chemical depletion of HD is negligible.
3.6. Deuterium abundance
To derive the deuterium abundance in the warm, shocked gas we adopt the non-LTE level distribution of HD we computed above, and further account for the possibility of chemical depletion. With the column density ratio derived for temperatures of 600– 900 K and densities of(1–3.5) × 105cm−3, we found
[D]/[H] = 0.5[HD]/[H2] = (4.5 ± 1.7) × 10−6, (13) and accounting for chemical depletion of HD by 1.67, the abun-dance is raised to
[D]/[H] = (7.6 ± 2.9) × 10−6, (14)
The main uncertainty of our value appears to arise from the indirect measure of the HD excitation, and of the abundance of atomic hydrogen in the warm shocked region. Less dissocia-tive shocks would require larger corrections for non-LTE level populations, which would raise the implied HD abundance: ne-glecting all H–HD collisions, we earlier derived N(HD) ≈
(3.5 ± 1.4) × 1016cm−2, which yields [D]/[H]= 7.9 × 10−6.
Now chemical depletion should not occur in shocks with a low abundance of neutral hydrogen, so that this values reflects the actual deuterium abundance. Interestingly, we find that the ef-fects of non-LTE and of chemical depletion nearly cancel, so that the derived deuterium abundance turns out to be not very sensitive on how dissociative the shock is.
4. Summary
ISO for the first time enabled the detection in the interstellar medium of an infrared transition in the electronic ground state of deuterated hydrogen, HD. We here report the discovery of thev=0–0 R(5) line at 19.4290 µm with the ISO Short Wave-length Spectrometer in the warm, shocked molecular gas of the Orion OMC-1 outflow, at the bright emission “Peak 1.” Up-per flux limits for sixteen other HD lines were measured, all of which appear consistent with expectations when considering the observed 0–0 R(5) line flux.
A large number of H2 lines were detected (Rosenthal et al. in prep.) and utilized to analyze the HD observations. The near- and mid-infrared extinction toward the emitting region was derived by minimizing the dispersion in the observed H2 level column densities with respect to an LTE excitation model. Thereby we derive a near-infrared (K band) extinction of(1.0±
0.1) magnitudes, a 9.7 µm extinction of (1.35±0.15) mag, and
from an estimated range ofA19/A9.7 = 0.35–0.52 we correct the HD 0-0 R(5) flux for extinction by(0.61 ± 0.15) mag, i.e. a factor1.75 ± 0.25.
The dereddened H2level populations served to estimate the excitation conditions in the gas. While H2was assumed to have thermalized level populations, those for HD were computed in detail by making use of the H2 excitation. Due to non-LTE effects atJ > 3, the total warm HD column density was found to be sensitive to the gas density and temperature at the densities estimated to prevail in the shocked gas,n(H2) ≈ (1–3.5) ×
105cm−3.
Our estimate for the observed warm HD column density is N(HD) = (2.0 ± 0.75) × 1016 cm−2, and for the warm molecular hydrogen,N(H2) = (2.21 ± 0.24) × 1021cm−2. Their relative abundance is therefore[HD]/[H2] = (9.0±3.5)×
10−6.
We note that in high-velocity C-shocks, HD may be depleted relative to H2because of an asymmetry (due to a small binding energy difference) in the deuterium-hydrogen exchange reac-tionHD + H ←→ D + H2. Estimating that this lowers the warm HD column by about 40%, we derive a deuterium abundance in the warm shocked gas, [D]/[H]= (7.6 ± 2.9) × 10−6.
If the emitting shocks were on average less dissociative than assumed, the chemical depletion would be less pronounced. But
at the same time, the lower H-HD collision rate would enhance the HDJ = 6 level population’s deviation from LTE, to the effect that our implied total column of HD would increase. We estimate that the two effects approximately cancel, and that [D]/[H] remains at≈ 8×10−6, independent of how dissociative the shock actually is.
The major uncertainties in our estimate for the deuterium abundance arise from our indirect measure of the HD excita-tion, and since we must therefore rely on non-LTE excitation models, on the uncertainty of the preshock density and the abun-dance of neutral hydrogen in the shock. Future ground-based near-IR observations of ro-vibrational transition lines of HD could constrain the effects of non-LTE and thereby narrow the error margins of the deuterium abundance. Detailed shock mod-els of the recently resolved, multiple bow-shaped emission in the Orion outflow (Schultz et al. 1998) might also yield better estimates for the pre-shock densities and shock velocities, and thereby of the H2 dissociation and of the chemical depletion fraction of HD in such shocks.
The deuterium abundance we find, (7.6 ± 2.9) × 10−6, is lower than that derived through most DI absorption mea-surements in the local ISM, but it is consistent with that found recently by Jenkins et al. (1999) toward λ Orionis,
(7.4+1.9
−1.3) × 10−6, and by Wright et al. (1999) toward the Orion
Bar,(1.0 ± 0.3) × 10−5. This could indicate that toward Orion, the deuterium abundance is indeed somewhat lower than on average.
Acknowledgements. We are thankful to J. Lacy, B. Draine, and
M. Walmsley for valuable comments, to A. Schultz for providing the NICMOS image, and to the SWS Data Center at MPE, especially to H. Feuchtgruber and E. Wieprecht, for their support. SWS and ISODC at MPE are supported by DARA under grants 50QI86108 and 50QI94023. FB acknowledges support by the Deutsche Forschungs-gemeinschaft (DFG) through its “Physics of Star Formation” program. CMW acknowledges support by NFRA/NWO grant 781-76-015 and of an ARC Research Fellowship.
References
Abgrall H., Roueff E., Viala Y., 1982, A&AS 50, 505
Beatty J.J., Garcia-Munoz J.E., Simpson J.A., 1985, ApJ 294, 455 Bertoldi F., 1997, In: Heras A.M., et al. (eds.) Proc. First ISO Workshop
on Analytical Spectroscopy. ESA SP-419, p. 67
Blake G.A., 1997, In: van Dishoeck E.F. (ed.) IAU Symposium 178, Molecules in Astrophysics: Probes and Processes. Kluwer, Dor-drecht, p. 31
Brand P.W.J.L., Moorhouse A., Burton M.G., et al., 1988, ApJ 334, L103
Burles S., Tytler D., 1998, ApJ 499, 699 Casuso E., Beckman J.E., 1997, ApJ 475, 155
Chengaluer J.N., Braun R., Burton W.B., 1997, A&A 318, L35 Chernoff D.F., Hollenbach D.J., McKee C.F., 1982, ApJ 259, L97 Chrysostomou A., Burton M.G., Axon D.J., et al., 1997, MNRAS 289,
605
de Graauw Th., Haser L., Beintema D.A., et al., 1996, A&A 315, L49 Draine B.T., 1989, In: Kaldeich B.H. (ed.) 22nd ESLAB Symposium,
Draine B.T., Roberge W.G., 1982, ApJ 259, L91 Draine B.T., Bertoldi F., 1996, ApJ 468, 269
Dring A.R., Linsky J., Murthy J., et al., 1997, ApJ 488, 760
Encrenaz Th., de Graauw Th., Schaeidt S., et al., 1996, A&A 315, L397 Essenwanger P., Gush H.P., 1984, Can. J. Phys. 62, 1680
Everette M.E., DePoy D.L., Pogge R.W., 1995, AJ 110, 1295 Feuchtgruber H., Lellouch E., de Graauw Th., 1997, BAAS DPS
meet-ing, 29, no. 15.16
Geiss J., Gloeckner G., 1998, Space Sci. Rev. 84, 239 Genzel R., Stutzki J., 1989, ARA&A 27, 41
Helmich F.P., van Dishoeck E.F., Jansen D.J., 1996, A&A 313, 657 Herzberg G., 1950, Molecular Spectra and Molecular Structure. Vol.1:
Spectra of diatomic molecules. Van Nostrand Reinhold, New York Jacq T., Walmsley,C.M., Mauersberger R., 1993, A&A 271, 276 Jenkins E.B., Tripp T.M., Wozniack P.R., Sofia U.J, Sonneborn G.,
1999, ApJ, submitted
Kaufman, M.J., Neufeld D.A., 1996, ApJ 456, 611
Kessler, M.F., Steinz J.A., Anderegg M.E., et al., 1996, A&A 1996, 315, L27
Linsky J.L., 1998, Space Sci. Rev. 84, 285
Linsky J.L. Brown A., Gayley K., et al., 1993, ApJ 402, 694 Linsky J.L., Diplas A., Wood B.E., et al., 1995, ApJ 451, 335 Lutz D., Feuchtgruber H., Genzel R., et al., 1997, A&A 315, L269 McCollough R.P., 1992, ApJ 390, 213
Menten K.M., Reid M.J., 1995, ApJ 445, L157 Mullan D.J., Linsky J.L., 1999, ApJ 511, 502 Neufeld D.A., Dalgarno A., 1989, ApJ 344, 251
Neufeld D.A., Melnick G.J., Harwit, M., 1998, ApJL, in press Pegourie B., Papoular R., 1985, A&A 142, 451
Piskunov N., Wood B.E., Linsky J.L., 1997, ApJ 474, 315 Roueff E., Flower D.R., 1999, MNRAS, in press Roueff E., Zeippen C.J., 1999, A&A, in press Sch¨afer J., 1990, A&AS 85, 1101
Schultz A.S.B, Colgan S.W.J., Erickson E.F., et al., 1998, ApJ, in press Smith M.S., Kawano L.H., Malaney R.A., 1993, ApJS 85, 219 Songaila A., Cowie L.L., Hogan C.J., et al., 1994, Nat 368, 599 Sternberg A., 1990, ApJ 361, 121
Stolovy S.R., Burton M.G., Erickson E.F., et al., 1998, ApJ 492, L151 Timmermann R., 1996, ApJ 456, 631
Timmermann R., 1998, ApJ 498, 246 Tosi M., 1998, Space Sci. Rev. 84, 207
Tosi M., Steigman G., Matteucci F., Chiappini C., 1998, ApJ 498, 226 Turner B.E., 1990, ApJ 362, L29
Turner J., Kirby-Docken K., Dalgarno A., 1977, ApJS 35, 281 Tytler D., Fan X.M., Burles S., 1996, Nat 381, 207
Ulivi L., De Natale P., Inguscio M., 1991, ApJ 378, L29
van Dishoeck E.F., Wright C.M., Cernicharo J., et al., 1998, ApJ 502, L173
Volk K., Kwok S., 1988, ApJ 331, 435
Wilson T.L., Rood R.T., 1994, ARA&A 32, 191
Wolniewicz L., Simbotin I., Dalgarno A., 1998, ApJS 115, 293 Wright C.M., van Dishoeck E.F., Cox P., Sidher S., Kessler M.F., 1999,
ApJ, submitted
